BigCapital beta

Article · Jobs and income

Labor productivity: how it grew and what drives it

Since Q1 1947, output per hour in US business has grown ×5.5 — 2.2% a year on average, but in spurts: 2.7% a year in 1947–1973, only 1.6% in 2004–2019. What a worker produced in an hour in 1947 now takes about 11 minutes.

Updated with Q3 2026 data14 min read

Output per hour since 1947
×5.5

2.2% a year, US nonfarm business, to Q3 2026

The hour of 1947 now takes
11 min

to produce what an hour of work produced in Q1 1947

Fastest era: 1995–2004
3.1%

a year; the slowest, 1973–1995: 1.5%

Share of growth from new ideas and methods (TFP), 1947–1973
65%

33% in 2004–2019, when growth slowed to 1.6% a year

Output per hour since 1973
+165%

hourly pay after inflation: +4% to +121%, depending on what you measure

What labor productivity is: the shrinking hour

Labor productivity is output per hour worked: everything a sector produces, after inflation, divided by the hours people spent producing it. In the US the Bureau of Labor Statistics measures it every quarter for the nonfarm business sector — most of the economy, leaving out government, nonprofits and farms.

Since Q1 1947 that number has grown ×5.5. Turn it around and the hour shrinks: what a worker produced in a full hour in 1947 now takes about 11 minutes. Over the last four quarters, to Q3 2026, output per hour rose 2.1%.

Minutes it takes to produce what one hour produced in Q1 1947
Q1 194760 minthe starting hour
Q4 197329 min×2.0 per hour
Q4 199521 min×2.8 per hour
Q4 200416 min×3.7 per hour
Q4 201913 min×4.7 per hour
Q3 202611 min×5.5 per hour
At the start of each era and in the latest quarter. Output per hour, nonfarm business sector, seasonally adjusted. Data: U.S. Bureau of Labor Statistics, Productivity and Costs, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain)

Five eras of growth

Productivity has never grown at a steady pace. In the golden age, 1947–1973, output per hour rose 2.7% a year and doubled every 26 years. Then came a long slowdown: 1.5% a year until 1995. The computer and internet boom lifted it back to 3.1% in 1995–2004; from 2004 to 2019 it sank to 1.6% again. Since 2019: 2.2% a year to Q3 2026, faster than the 1.6% of 2004–2019.

Why computers took so long to show up in these numbers is the famous productivity paradox — it has its own chapter in Economic paradoxes. Here the question is what moved the numbers in each era; the next section splits the growth into its sources.

Era: growth a yearSingle years (stronger blue = faster, red = a fall)Average since 1947: 2.2%
Each block spans its era; its height is the era's average growth a year (compound, quarter to quarter). The strip: each calendar year's growth, annual average to annual average. Data: U.S. Bureau of Labor Statistics, Productivity and Costs, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain)
EraYearsA yearOver the eraDoubles every
Golden ageQ1 1947 – Q4 19731947–19732.7%×2.0526 years
The slowdownQ4 1973 – Q4 19951973–19951.5%×1.3946 years
IT boomQ4 1995 – Q4 20041995–20043.1%×1.3223 years
The second slowdownQ4 2004 – Q4 20192004–20191.6%×1.2645 years
The 2020sQ4 2019 – Q3 20262019–20262.2%×1.1632 years

The road not taken

Small differences compound. Had output per hour kept its golden-age pace of 2.7% a year after 1973, it would now be 1.5 times its actual level. The gap between the dashed line and the real one is the price of the slowdowns.

Output per hourAt the 1947–1973 pace of 2.7% a year
Output per hour, nonfarm business sector, Q1 1947 = 1. Log scale: equal steps are equal percentage changes. Data: U.S. Bureau of Labor Statistics, Productivity and Costs, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain)

What drives it: capital, skills and ideas

Economists split productivity growth into three sources. Capital deepening: more and better equipment, buildings, software and research for every hour worked. Labor composition: a workforce with more education and experience. And total factor productivity (TFP): everything else — new technologies and ideas, better management and organization, work moving to more productive firms. TFP is what is left once capital and skills are counted, so it also collects measurement errors.

In the golden age TFP added 2.01 of the 3.10 percentage points of growth a year (65%). It fell to 0.52 in 1973–1995, came back to 1.81 in the IT boom and dropped to 0.49 in 2004–2019; since 2019 it has added 1.16. Capital deepening ranged from 0.71 to 1.25 points a year across the eras, skills from 0.19 to 0.31.

So the slowdowns were above all slowdowns of TFP: of the 1.62 points a year that growth lost between 1947–1973 and 2004–2019, TFP accounts for 94%.

Capital deepeningLabor composition (skills)Total factor productivity (TFP)
Average contribution to the growth of output per hour a year, percentage points (log changes at an annual rate), Q2 1947 – Q3 2026. Business sector including farms, so the totals differ a little from the nonfarm figures above. Data: John Fernald, Federal Reserve Bank of San Francisco, quarterly growth accounts of the US business sector (from BEA and BLS data)

BLS's own estimates for the private nonfarm business sector, on FRED from 1988, tell the same story: TFP added 1.50 points a year in 1995–2004 and 0.60 in 2004–2019.

The split as a table
EraCapital deepeningLabor compositionTFPOutput per hourTFP's share
Golden ageQ1 1947 – Q4 19730.900.192.013.1065%
The slowdownQ4 1973 – Q4 19950.710.310.521.5334%
IT boomQ4 1995 – Q4 20041.250.241.813.3055%
The second slowdownQ4 2004 – Q4 20190.730.250.491.4833%
The 2020sQ4 2019 – Q3 20260.960.231.162.3549%
BLS's own split, since 1988
PeriodOutput per hourCapital deepeningLabor compositionTFP
1987–1995*1.560.670.390.49
1995–20043.131.320.261.50
2004–20191.580.690.280.60
2019–20252.230.850.331.03

Private nonfarm business sector, averages of the yearly changes (percent and percentage points). BLS publishes each figure to one decimal, so the parts may not add up exactly. * FRED carries these series from 1988; the slowdown began in 1973. Data: U.S. Bureau of Labor Statistics, Total Factor Productivity for Major Sectors, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain)

Who produces the most per hour

Comparing countries needs a common yardstick: output valued at purchasing-power parities — what money buys locally — divided by the hours worked. By that measure the US produced $84 an hour in 2023, No. 8 of the 126 countries with at least 1 million people and hours data. The top three were Ireland (189% of the US level), Norway and Switzerland. Ireland's figure is lifted by the profits of multinational companies booked there. Norway's includes oil and gas, which take few hours to produce a lot of value. Among large economies, with more than 30 million people, the leaders were the United States (100%), Germany (99%) and France (93%).

Hours matter as much as output. In 2023 a worker in Germany put in 1,335 hours on average, against 1,789 in the US, and produced about as much per hour (99% of the US level) — but only 74% as much per worker. A shorter working year is productivity taken as free time instead of income.

Output per hour worked in 2023
#CountryOutput per hour% of USHours a year per worker
1Ireland$158189%1,654
2Norway$130156%1,412
3Switzerland$100120%1,529
4Denmark$89106%1,380
5Austria$85102%1,438
6Netherlands$85102%1,439
7Belgium$84101%1,593
8United States$84100%1,789
9Germany$8399%1,335
10Sweden$8197%1,432
11France$7893%1,487
12Australia$7792%1,611
Other large economies (among the 10 largest)
15United Kingdom$6881%1,523
30Japan$4959%1,654
49Russia$3744%2,087
69Brazil$2024%1,994
77China$1821%2,328
86Indonesia$1417%1,958
100India$810%2,383

Real GDP (output side, at chained purchasing-power parities, 2021 US dollars) per hour worked by everyone engaged in production; countries with at least 1 million people and hours data. The ranks of small, specialized economies move a lot with commodity prices and company accounting. Data: Penn World Table 11.0 (Feenstra, Inklaar and Timmer, Groningen Growth and Development Centre), CC BY 4.0; figures calculated by BigCapital

Catching up with the US — and falling back

Each panel shows a country's output per hour as a share of the US level in the same year. 2 of these 12 countries produced as much per hour as the US or more in 2023: the Netherlands and Ireland. Ireland climbed the most, from 24% of the US level in 1950 to 189%; Mexico fell back the most, from a peak of 78% in 1981 to 32% in 2023.

Output per hour, % of the US levelUS = 100
Real GDP per hour worked at chained purchasing-power parities, as a share of the US figure of the same year, 1950–2023; a line starts when the country's hours data do. Data: Penn World Table 11.0 (Feenstra, Inklaar and Timmer, Groningen Growth and Development Centre), CC BY 4.0; figures calculated by BigCapital
The panels as a table (US = 100)
Country19501960197019801990200020102023Peak
Germany25405167829897991042019
France364665899410194931032019
Netherlands5963751041001081041021172008
United Kingdom575966798310585811052001
Italy2734548486938081951998
Spain2230466669767577812019
Japan1620355166767059791996
South Korea—781429435564662022
Taiwan—14233656746873742005
Canada6779908990898074931984
Ireland2427294455851161892142022
Mexico5465747650444032781981

Do poorer countries catch up?

Economic theory says they should: a poorer country can copy technologies instead of inventing them, and each new machine adds more where there are few. The scatter tests it on 107 countries with at least 1 million people: output per worker in 1970, as a share of the US level, against how fast output per worker grew from 1970 to 2023.

The pull exists, but it is not automatic. Of the 56 countries that started below 25% of the US level, 30 grew faster than the US (1.5% a year) and closed part of the gap; 26 fell further behind. The fastest were China (7.0% a year, from 2% of the US level), Taiwan (4.5%) and South Korea (4.4%).

Among the 27 countries that started above 50% of the US level, catch-up came close to a rule: the further behind, the faster they grew (correlation −0.75). The United Arab Emirates, Bahrain, Kuwait, Qatar and Saudi Arabia started out ahead of the US per worker in 1970 and have produced less per worker since: oil exporters, whose output follows the price of oil.

Grew faster than the US (1.5% a year)Grew slower
Each dot is a country; its size, the population in 2023. Across: output per worker in 1970, % of the US level, log scale. Up: growth of output per worker a year, 1970–2023, at constant national prices. Lines: the US. Data: Penn World Table 11.0 (Feenstra, Inklaar and Timmer, Groningen Growth and Development Centre), CC BY 4.0; figures calculated by BigCapital
The fastest and the slowest
Fastest growth, 1970–2023
Country1970, % of USA year
China2%+7.0%
Taiwan27%+4.5%
South Korea12%+4.4%
Myanmar3%+4.4%
Mali3%+4.4%
Vietnam4%+3.9%
Botswana4%+3.8%
Romania10%+3.7%
Ireland36%+3.5%
Bulgaria16%+3.3%
Slowest growth, 1970–2023
Country1970, % of USA year
Venezuela61%−3.2%
Kuwait658%−3.2%
Saudi Arabia325%−2.5%
United Arab Emirates903%−2.2%
Qatar390%−1.5%
Madagascar6%−1.4%
DR Congo11%−1.2%
Lebanon43%−1.1%
Bahrain131%−1.1%
Oman80%−0.7%

Productivity and pay

Does higher productivity raise pay? Over the long run it has to — wages are paid out of output. But how closely pay has followed productivity since 1973 is one of the most argued-over charts in economics, and the answer turns on measurement choices. The staircase changes one choice at a time, from 1973 to Q3 2026.

Growth since 1973, after inflation
  1. Output per hournonfarm business+165%
  2. −44 ptslabor's share of income fell 17%
  3. Hourly pay, all workerswages + benefits, in the prices of what businesses produce+121%
  4. −27 ptsconsumer prices rose faster than the prices of business output
  5. Hourly pay, all workersin consumer prices: the PCE price index+95%
  6. −43 ptsthe CPI rose faster than the PCE price index
  7. Hourly pay, all workersin consumer prices: the CPI+52%
  8. −47 ptsbenefits and the pay of other workers grew faster than rank-and-file wages
  9. Hourly wages, rank-and-file workersproduction and nonsupervisory workers, wages only, in CPI prices+4%
Change from the 1973 average to Q3 2026. Each step changes one measurement choice; the gap between two steps says what changed. Data: U.S. Bureau of Labor Statistics, Productivity and Costs, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain) Data: U.S. Bureau of Economic Analysis, PCE price index, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain)

Output per hour rose 165%. The hourly pay of all workers — wages plus benefits — measured in the prices of what businesses produce rose 121%; the difference is the change in labor's share of income (−17%), and why it moved is one of the open questions in Economic mysteries. Measured in what workers buy, the same pay rose 95% with the PCE price index and rose 52% with the CPI. The hourly wages of production and nonsupervisory workers, without benefits, rose 4% after CPI inflation.

Each step answers a different question. The first two ask whether workers as a whole receive their share of what they produce; the last ones ask what an ordinary paycheck buys. The choice of price index is no detail: over these years the CPI rose faster than the PCE index, partly because of how it is built — a fixed basket and, before 1983, the way it counted home prices.

Output per hourPay, output pricesPay, PCE pricesPay, CPI pricesRank-and-file wages, CPI
Index, 1973 average = 100, quarterly. Pay: hourly compensation of all workers in the nonfarm business sector. Wages: average hourly earnings of production and nonsupervisory employees, private sector. Data: U.S. Bureau of Labor Statistics, Productivity and Costs, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain) Data: U.S. Bureau of Economic Analysis, PCE price index, retrieved from FRED, Federal Reserve Bank of St. Louis (public domain)

Era by era, pay kept to productivity's rhythm: in the two fastest eras, 1995–2004 and 1947–1973, pay in output prices grew 3.0% and 2.6% a year; in the two slowest, 2004–2019 and 1973–1995, 1.1% and 1.2%.

Growth a year by era
EraOutput per hourPay, output pricesPay, CPI prices
Golden ageQ1 1947 – Q4 19732.7%2.6%2.5%
The slowdownQ4 1973 – Q4 19951.5%1.2%0.3%
IT boomQ4 1995 – Q4 20043.1%3.0%2.0%
The second slowdownQ4 2004 – Q4 20191.6%1.1%0.7%
The 2020sQ4 2019 – Q3 20262.2%1.3%1.1%

Questions people ask

What is labor productivity?

Output per hour worked: the real value of what is produced, divided by the hours it took. In the US, BLS measures it each quarter for the nonfarm business sector; it has grown ×5.5 since Q1 1947 — 2.2% a year — and rose 2.1% over the last four quarters, to Q3 2026.

Why has productivity growth slowed?

Mainly because total factor productivity — new technology, ideas and better organization — slowed: it added 2.01 percentage points a year in 1947–1973 and 0.49 in 2004–2019, while capital per hour added 0.90 and 0.73. Why TFP slowed is still debated: fewer world-changing inventions, slower spread of technology from leading firms to the rest, and output that statistics struggle to measure, such as free digital services.

What increases productivity?

By the usual accounting, three things: more capital per hour — equipment, software, research — which added 0.86 percentage points a year on average from Q2 1947 to Q3 2026; a more educated and experienced workforce, 0.24; and total factor productivity, 1.22 — the largest part, 52% of the 2.32 total.

Which country has the highest productivity?

Per hour worked in 2023: Ireland ($158 an hour), Norway ($130) and Switzerland ($100) led the countries with at least 1 million people; the US produced $84, No. 8 of 126. Among economies with more than 30 million people the leaders were the United States (100%), Germany (99%) and France (93%).

Does productivity raise wages?

Over long periods, yes: eras of fast productivity growth have also been eras of fast pay growth. Since 1973, output per hour rose 165% and the hourly pay of all workers rose 121% in the prices of what they produce. Rank-and-file hourly wages did much worse: they rose 4% after CPI inflation, because of the gaps in the staircase above: labor's share of income, consumer prices against output prices, and wages against total pay.

Go deeper

How the numbers are calculated

  • Output per hour, hourly compensation and the output price deflator of the nonfarm business sector, and average hourly earnings of production and nonsupervisory employees: U.S. Bureau of Labor Statistics (series OPHNFB, COMPNFB, IPDNBS, AHETPI); consumer prices: BLS CPI-U (CPIAUCSL) and BEA's PCE price index (PCEPI); all retrieved from FRED, Federal Reserve Bank of St. Louis. US government works, public domain.
  • The split of growth into capital, labor composition and TFP: John G. Fernald, “A Quarterly, Utilization-Adjusted Series on Total Factor Productivity”, FRBSF Working Paper 2012-19, Federal Reserve Bank of San Francisco (file of November 06, 2026); business sector, from BEA and BLS data; reproduced with credit to the Federal Reserve Bank of San Francisco. BLS's own split (private nonfarm business, from 1988): series MPU4910063, MPU4910153, MPU4910183 and MPU4910013, via FRED.
  • Countries: Penn World Table 11.0, Groningen Growth and Development Centre, licensed under CC BY 4.0. Please cite: Feenstra, Robert C., Robert Inklaar and Marcel P. Timmer (2015), "The Next Generation of the Penn World Table" American Economic Review, 105(10), 3150-3182, available for download at www.ggdc.net/pwt. Figures calculated by BigCapital.
  • Eras run between the quarters Q1 1947, Q4 1973, Q4 1995, Q4 2004, Q4 2019 and Q3 2026, the latest (the same cut as in Economic paradoxes); growth a year is compound between those quarters. Single years compare annual averages; the year under way is left out.
  • Contributions: capital deepening = capital's share of income × growth of capital input per hour; labor composition = labor's share × growth of labor composition; TFP = output growth minus both inputs. Averages of quarterly log changes at an annual rate, so the three add up to the growth of output per hour exactly.
  • Countries: output per hour = output-side real GDP at chained PPPs (rgdpo) ÷ (persons engaged × average annual hours); output per worker = rgdpo ÷ persons engaged; its growth uses real GDP at constant national prices (rgdpna). The ranking and the scatter include countries with at least 1 million people in 2023.
  • Pay: indexes with the 1973 average = 100; monthly prices and earnings are averaged into quarters. Pay in output prices = hourly compensation ÷ the nonfarm business output price deflator; its ratio to output per hour is labor's share of income (checked against BLS's labor share series). Quarters with one month missing in a monthly series are averaged over the other two: Q4 2025. BLS published no CPI for Oct 2025.
  • History, not a forecast. Nothing here is investment advice.

Reuse this article

The text and charts are free to republish, translate and adapt under CC BY 4.0: credit BigCapital.org with a link to this page. Data from other providers keeps its own license (see the sources below). How to credit

Source: BigCapital.org — “Labor productivity”, https://bigcapital.org/articles/labor-productivity/